mitchellhansen/voxel-raycaster
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

MBP was having problems with out of bounds memory operations with the way the cam dir was handled. sf::vector3f -> float3 and while accessign the Zth element. I'm assuming it was because of some weird backend stuff regarding that gentypeOdds are actually gentypeOdds + 1. Converted write_imagef's to write_imageui's though I don't think that really helps anything. Fixed the bottom half of the screen getting cut off. View matrix import error. Fixed problem the MBP had with negative values during device init, that was a weird one.

Browse Source
This commit is contained in:
mitchellhansen
2016-09-05 22:01:07 -07:00
parent 4e96985104
commit 10bc771807
4 changed files with 59 additions and 40 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
src/CL_Wrapper.cpp
View File

@@ -59,8 +59,9 @@ int CL_Wrapper::acquire_platform_and_device(){
// falling back to the cpu with the fastest clock if we weren't able to find one
device current_best_device;
current_best_device.type = -1; // Set this to -1 so the first run always selects a new device
current_best_device.type = 0; // Set this to -1 so the first run always selects a new device
current_best_device.clock_frequency = 0;
current_best_device.comp_units = 0;
for (auto kvp: plt_ids){
@@ -86,7 +87,7 @@ int CL_Wrapper::acquire_platform_and_device(){
platform_id = current_best_device.platform;
device_id = current_best_device.id;
return 0;
return 1;
};
int CL_Wrapper::create_shared_context() {
@@ -240,7 +241,7 @@ int CL_Wrapper::store_buffer(cl_mem buffer, std::string buffer_name){
int CL_Wrapper::run_kernel(std::string kernel_name, const int work_size){
const int WORKER_SIZE = 10;
const int WORKER_SIZE = 1;
size_t global_work_size[1] = { static_cast<size_t>(work_size) };
cl_kernel kernel = kernel_map.at(kernel_name);